Elsevier

Clinical Radiology

Volume 66, Issue 5, May 2011, Pages 434-439
Clinical Radiology

Original Paper
The use of ultrasound to diagnose hepatic steatosis in type 2 diabetes: Intra- and interobserver variability and comparison with magnetic resonance spectroscopy

https://doi.org/10.1016/j.crad.2010.09.021Get rights and content

Aim

To compare ultrasound gradings of steatosis with fat fraction (FF) on magnetic resonance spectroscopy (MRS; the non-invasive reference standard for quantification of hepatic steatosis), and evaluate inter- and intraobserver variability in the ultrasound gradings.

Materials and methods

Triple grading of hepatic ultrasound examination was performed by three independent graders on 131 people with type 2 diabetes. The stored images of 60 of these individuals were assessed twice by each grader on separate occasions. Fifty-eight patients were pre-selected on the basis of ultrasound grading (normal, indeterminate/mild steatosis, or severe steatosis) to undergo 1H-MRS. The sensitivity and specificity of the ultrasound gradings were determined with reference to MRS data, using two cut-offs of FF to define steatosis, ≥9% and ≥6.1%.

Results

Median (intraquartile range) MRS FF (%) in the participants graded on ultrasound as normal, indeterminate/mild steatosis, and severe steatosis were 4.2 (1.2–5.7), 4.1 (3.1–8.5) and 19.4 (12.9–27.5), respectively. Using a liver FF of ≥6.1% on MRS to denote hepatic steatosis, the unadjusted sensitivity and specificity of ultrasound gradings (severe versus other grades of steatosis) were 71 and 100%, respectively. Interobserver agreement within one grade was observed in 79% of cases. Exact intraobserver agreement ranged from 62 to 87%.

Conclusion

Hepatic ultrasound provided a good measure of the presence of significant hepatic steatosis with good intra- and interobserver agreement. The grading of a mildly steatotic liver was less secure and, in particular, there was considerable overlap in hepatic FF with those who had a normal liver on ultrasound.

Introduction

Non-alcoholic fatty liver disease (NAFLD), i.e., hepatic steatosis in the absence of a secondary cause, is a common cause of chronic liver disease. It is more common in people with type 2 diabetes than in the general population, and studies have suggested that it may affect up to 75% of individuals in this group.1, 2, 3 It is important to detect this condition as it can result in considerable morbidity and mortality,4, 5, 6, 7 and may independently predict the risk of cardiovascular events.8 Its identification can be used to initiate screening and target treatment for hepatic cirrhosis.

Ultrasound imaging is an established imaging modality in the diagnosis of hepatic steatosis, both clinically and in large-scale studies. Although several grading systems have been proposed for the assessment of hepatic steatosis using ultrasound, no consensus has been achieved.3, 9, 10, 11, 12 Detection of hepatic fat is based on well-established characteristics including an echogenic parenchyma (especially in relation to the right kidney), posterior attenuation, and areas of focal fatty sparing.3, 11, 12, 13, 14 Compared with liver biopsy, it has been reported to have a sensitivity of 83–100% and specificity of 84–100% in the diagnosis of hepatic steatosis.9, 11, 13, 15 It also has the advantages over biopsy of safety, ease of acquisition, and ability to assess the whole liver.

Although a degree of subjectivity is recognized when diagnosing hepatic steatosis by ultrasound examination, few studies have examined intra- and interobserver variability in making this diagnosis. One study of 25 patients with biopsy-proven NAFLD (less than a quarter of whom had diabetes) reported that the interobserver correlation between two radiologists in determining severity of steatosis was 0.40 (representing fair agreement), and the intraobserver correlation was 0.63 (suggesting more substantial agreement).9 A different study comprising 168 patients who had undergone abdominal ultrasonography for a variety of abdominal disorders, demonstrated interobserver agreement between three radiologists regarding the presence and severity of hepatic steatosis of 0.40–0.51 and intraobserver agreement of 0.58 (moderate agreement).16 In both studies, the assessment of the severity of steatosis was based on the presence and severity of increased echogenicity of the hepatic parenchyma and attenuation of the ultrasound beam.

In comparison with ultrasound imaging, magnetic resonance spectroscopy (MRS), the non-invasive reference standard for measurement of hepatic fat, is expensive but has a high correlation with liver fat concentration on biopsy, of the order of 0.7–0.9.17, 18 Different “normal” values for hepatic fat fraction (FF) on magnetic resonance imaging have been reported previously.19, 20, 21 The Dallas Heart Study used MRS to determine hepatic FF in a population of 345 participants who had normal body mass index (BMI), glucose tolerance, and liver function tests, and non-excessive alcohol use (i.e., a very low risk for developing hepatic steatosis). In this group the 95th percentile of hepatic triglyceride content (expressed as grams of triglyceride per 100 g wet liver tissue) was 5.5%, corresponding to an MRS fat fraction of 6.1%,17 and this was taken as the upper limit of normal.20 In an earlier smaller study, 28 healthy volunteers underwent magnetic resonance imaging (MRI) using a modified Dixon gradient echo technique. All had a magnetic resonance hepatic FF under 9% and this was later considered as being the upper limit of normal.19, 21 In the same study seven patients with cystic fibrosis underwent both MRI and liver biopsy: two patients who had no steatosis on biopsy had a FF on MRI of <9%; five patients who had mild to severe steatosis had a FF of >9%. A further study found that the geometric mean intrahepatic lipid content (percentage ratio of CH2 lipid peak area relative to the water peak area) in 23 healthy volunteers to be 2.7 (range 0.2–77.4).22 There is a paucity of data comparing qualitative ultrasound gradings with FF on MRS.

The aim of the present study was to compare ultrasound gradings of hepatic steatosis, performed by three graders, with hepatic FF on MRS, and also to examine inter- and intraobserver variability in these ultrasound gradings.

Section snippets

Materials and methods

The selection of participants for the Edinburgh Type 2 Diabetes Study has been previously described in detail. Briefly, patients recorded as having type 2 diabetes, aged 60–74 years, were selected at random into sex and 5 year age bands from the Lothian Diabetes Register, a computerized database that contains details of over 20,000 patients with type 2 diabetes living in Lothian, Scotland. Invitations to participate in an initial clinic were sent to 5454 people and, of these, 1077 (19%)

MRS subgroup

All 58 participants with spectroscopy measurements underwent the repeat ultrasound examination (ultrasound2) and this was graded by three graders; the number of participants put into each category for each grader is listed in Table 1. For grader 1, median MRS FF was 4.2% (IQ range 1.2–5.7%) in the group graded normal, 4.1% (IQ range 3.1–8.5%) in the group graded indeterminate/mildly steatotic, and 19.4% (IQ range 12.9–27.5%) in the group graded severely steatotic (Fig 1). There was a

Discussion

To the authors’ knowledge this is the first study to examine both the relationship between ultrasound gradings of steatosis and MRS, and the inter- and intraobserver agreement in ultrasound grading. The present study was broad in its scope, including both participants who were likely to have a normal liver and those considered to have varying degrees of hepatic steatosis, mainly NAFLD. In addition, the use of ultrasound has been examined in diagnosing hepatic steatosis in people with type 2

Acknowledgements

This study was funded by Pfizer Ltd and the UK Medical Research Council. Research clinics were performed in the Wellcome Trust Clinical Research Facility and MRS carried out in the Scottish Funding Council Brain Imaging Centre, Western General Hospital, Edinburgh.

References (22)

  • P. Angulo et al.

    Independent predictors of liver fibrosis in patients with nonalcoholic steatohepatitis

    Hepatology

    (1999)
  • Cited by (0)

    View full text